Visiting any page on github.com with 2.13.92 result in segmentation fault once the page is mostly rendered. Build arguments cmake -DPORT=GTK -DCMAKE_BUILD_TYPE=Release \ -DCMAKE_SKIP_RPATH=ON -DCMAKE_INSTALL_PREFIX=/usr \ -DLIB_INSTALL_DIR=/usr/lib -DLIBEXEC_INSTALL_DIR=/usr/lib/webkit2gtk-4.0 \ -DENABLE_GTKDOC=OFF -DENABLE_MEDIA_SOURCE=ON -DENABLE_VIDEO_TRACK=ON -DENABLE_VIDEO=ON
We can't reproduce it, got a backtrace?
It crashes whole browser on Wayland, but only crashes tab on X11. locale LANG=C LC_CTYPE="C" LC_NUMERIC="C" LC_TIME="C" LC_COLLATE="C" LC_MONETARY="C" LC_MESSAGES="C" LC_PAPER="C" LC_NAME="C" LC_ADDRESS="C" LC_TELEPHONE="C" LC_MEASUREMENT="C" LC_IDENTIFICATION="C" LC_ALL=C Process 1817 (WebKitWebProces) dumped core. Stack trace of thread 1817: #0 0x00007f9f6577b4b1 _ZN3JSC18IntlDateTimeFormat15resolvedOptionsERNS_9ExecStateE (libjavascriptcoregtk-4.0.so.18) #1 0x00007f9f6577ed09 _ZN3JSCL46IntlDateTimeFormatPrototypeFuncResolvedOptionsEPNS_9ExecStateE (libjavascriptcoregtk-4.0.so. #2 0x00007f9f03fff028 n/a (n/a) Sep 19 23:07:05 systemd-coredump[1874]: Process 1796 (epiphany) dumped core. Stack trace of thread 1796: #0 0x00007f2d0bcac23e _ZNK6WebKit17WaylandCompositor6Buffer4sizeEv (libwebkit2gtk-4.0.so.37) #1 0x00007f2d0bcac913 _ZN6WebKit17WaylandCompositor7Surface25prepareTextureForPaintingERjRN7WebCore7IntSizeE (libwebkit2gtk- #2 0x00007f2d0bca4f65 _ZN6WebKit30AcceleratedBackingStoreWayland5paintEP6_cairoRKN7WebCore7IntRectE (libwebkit2gtk-4.0.so.37 #3 0x00007f2d0bc8c114 _ZL21webkitWebViewBaseDrawP10_GtkWidgetP6_cairo (libwebkit2gtk-4.0.so.37) #4 0x00007f2d0af910ab n/a (libgtk-3.so.0) #5 0x00007f2d0ad84260 gtk_container_propagate_draw (libgtk-3.so.0) #6 0x00007f2d0ad84342 n/a (libgtk-3.so.0) #7 0x00007f2d0af910ab n/a (libgtk-3.so.0) #8 0x00007f2d0ad84260 gtk_container_propagate_draw (libgtk-3.so.0) #9 0x00007f2d0ae7c200 n/a (libgtk-3.so.0) #10 0x00007f2d0ad8930d n/a (libgtk-3.so.0) #11 0x00007f2d0ad8df70 n/a (libgtk-3.so.0) #12 0x00007f2d0ae7c011 n/a (libgtk-3.so.0) #13 0x00007f2d0af910ab n/a (libgtk-3.so.0) #14 0x00007f2d0ad84260 gtk_container_propagate_draw (libgtk-3.so.0) #15 0x00007f2d0ad84342 n/a (libgtk-3.so.0) #16 0x00007f2d0ad38fa4 n/a (libgtk-3.so.0) #17 0x00007f2d0ad8930d n/a (libgtk-3.so.0) #18 0x00007f2d0ad8df70 n/a (libgtk-3.so.0) #19 0x00007f2d0ad3b941 n/a (libgtk-3.so.0) #20 0x00007f2d0af910ab n/a (libgtk-3.so.0) #21 0x00007f2d0ad84260 gtk_container_propagate_draw (libgtk-3.so.0) #22 0x00007f2d0ae6e2ba n/a (libgtk-3.so.0) #23 0x00007f2d0ad8930d n/a (libgtk-3.so.0) #24 0x00007f2d0ad8df70 n/a (libgtk-3.so.0) #25 0x00007f2d0ad3d074 n/a (libgtk-3.so.0) #26 0x00007f2d0ad8df70 n/a (libgtk-3.so.0) #27 0x00007f2d0ae70a0c n/a (libgtk-3.so.0) #28 0x00007f2d0af910ab n/a (libgtk-3.so.0) #29 0x00007f2d0ad84260 gtk_container_propagate_draw (libgtk-3.so.0) #30 0x00007f2d0ad84342 n/a (libgtk-3.so.0) #31 0x00007f2d0af9ea11 n/a (libgtk-3.so.0) #32 0x00007f2d0af910ab n/a (libgtk-3.so.0) #33 0x00007f2d0af91589 gtk_widget_send_expose (libgtk-3.so.0) #34 0x00007f2d0ae43925 gtk_main_do_event (libgtk-3.so.0) #35 0x00007f2d0a97a785 n/a (libgdk-3.so.0) #36 0x00007f2d0a989478 n/a (libgdk-3.so.0) #37 0x00007f2d0a98a5cc n/a (libgdk-3.so.0) #38 0x00007f2d0a98a783 n/a (libgdk-3.so.0) #39 0x00007f2d092ca1d4 n/a (libgobject-2.0.so.0) #40 0x00007f2d092e490d g_signal_emit_valist (libgobject-2.0.so.0) #41 0x00007f2d092e4fff g_signal_emit (libgobject-2.0.so.0) #42 0x00007f2d0a982711 n/a (libgdk-3.so.0) #43 0x00007f2d0a970ae8 n/a (libgdk-3.so.0) #44 0x00007f2d08112703 n/a (libglib-2.0.so.0) #45 0x00007f2d08111c8a g_main_context_dispatch (libglib-2.0.so.0) #46 0x00007f2d08112040 n/a (libglib-2.0.so.0) #47 0x00007f2d081120ec g_main_context_iteration (libglib-2.0.so.0) #48 0x00007f2d095b9d2d g_application_run (libgio-2.0.so.0) #49 0x00000000004023eb main (epiphany) #50 0x00007f2d07d4a291 __libc_start_main (libc.so.6) #51 0x000000000040290a _start (epiphany)
Please, use 'coredumpctl gdb' and then 'bt full' to get a good backtrace. You also need to install debuginfo for GLib and GTK+, or build them yourself with debuginfo if Arch doesn't provide any. Since it's crashing the UI process under Wayland, it would be interesting to see a backtrace of that process as well; a web process crash should never crash the UI process, so there are multiple bugs here.
It crashes if locale is set to C. There is no crash if I start using LC_ALL=en_US.UTF-8 epiphany. And adding a check in IntlDateTimeFormat also fixes the issue. if(!m_locale.isEmpty() ) { options->putDirect(vm, vm.propertyNames->locale, jsNontrivialString(&exec, m_locale)); }
OK, most likely a regression from "[INTL] Implement Intl.DateTimeFormat.prototype.resolvedOptions ()"
Created attachment 289442 [details] Patch
Thanks for the patch. If this patch contains new public API please make sure it follows the guidelines for new WebKit2 GTK+ API. See http://trac.webkit.org/wiki/WebKitGTK/AddingNewWebKit2API
Comment on attachment 289442 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=289442&action=review > LayoutTests/js/intl-invalid-locale-crash.html:10 > +if (window.internals) { > + window.internals.setUserPreferredLanguages(["a"]); Please add a comment to indicate that you're doing this to ensure the script gets run with an invalid locale. > Source/JavaScriptCore/runtime/IntlCollator.cpp:261 > + throwTypeError(&state, scope, ASCIILiteral("failed to initialize Collator")); Wouldn't it be good to indicate why? "failed to initialize Collator due to invalid locale"? > Source/WebKit2/UIProcess/API/gtk/WebKitWebContext.cpp:976 > + for (size_t i = 0; languageList[i]; ++i) { > + // Dot not propagate the C locale to WebCore. "Do not" And since you changed the line above anyway, I would change size_t to auto as well. > Source/WebKit2/UIProcess/API/gtk/WebKitWebContext.cpp:-979 > - WebCore::languageDidChange(); What, why dropping this? This looks like an accident. Please don't commit without responding to this. > Tools/TestWebKitAPI/Tests/WebKit2Gtk/TestWebKitWebContext.cpp:491 > + // When using the C locale, en-US should be used as default. I would copy/paste this entire block to add a test for POSIX locale as well.
(In reply to comment #8) > Comment on attachment 289442 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=289442&action=review > > > LayoutTests/js/intl-invalid-locale-crash.html:10 > > +if (window.internals) { > > + window.internals.setUserPreferredLanguages(["a"]); > > Please add a comment to indicate that you're doing this to ensure the script > gets run with an invalid locale. Ah, ok, I assumed the test name was enough. > > Source/JavaScriptCore/runtime/IntlCollator.cpp:261 > > + throwTypeError(&state, scope, ASCIILiteral("failed to initialize Collator")); > > Wouldn't it be good to indicate why? "failed to initialize Collator due to > invalid locale"? Ok. > > Source/WebKit2/UIProcess/API/gtk/WebKitWebContext.cpp:976 > > + for (size_t i = 0; languageList[i]; ++i) { > > + // Dot not propagate the C locale to WebCore. > > "Do not" Oops. > And since you changed the line above anyway, I would change size_t to auto > as well. What would auto deduce for 0? int? I prefer to be explicit here, we want an unsigned type. > > Source/WebKit2/UIProcess/API/gtk/WebKitWebContext.cpp:-979 > > - WebCore::languageDidChange(); > > What, why dropping this? This looks like an accident. Please don't commit > without responding to this. No, but I forgot to comment about this in the changelog, WebCore::overrideUserPreferredLanguages() already calls languageDidChange()), so we were calling all the observers twice. > > Tools/TestWebKitAPI/Tests/WebKit2Gtk/TestWebKitWebContext.cpp:491 > > + // When using the C locale, en-US should be used as default. > > I would copy/paste this entire block to add a test for POSIX locale as well. Good point.
I'll submit a new patch for landing, but I would like to hear from any JSC developer before, could someone confirm this approach is ok?
Created attachment 289551 [details] Patch for landing
(In reply to comment #9) > What would auto deduce for 0? int? I prefer to be explicit here, we want an > unsigned type. Oops, probably
Committed r206295: <http://trac.webkit.org/changeset/206295>